Current Biology 21, 154–160, January 25, 2011 ª2011 Elsevier Ltd All rights reserved

DOI 10.1016/j.cub.2010.12.038

Report A Nonmotor Microtubule Binding Site in Kinesin-5 Is Required for Filament Crosslinking and Sliding
Joshua S. Weinger,1 Minhua Qiu,2 Ge Yang,2 and Tarun M. Kapoor1,* 1Laboratory of Chemistry and Cell Biology, The Rockefeller University, New York, NY 10065, USA 2Department of Biomedical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA disruptive based on studies of a tagged full-length construct, Kin5-GFP [8, 9]. Comparison with Kin5-GFP by gel filtration chromatography and single-particle fluorescence intensity analysis indicated that Kin5-Dmotor-GFP is homotetrameric, like the full-length construct (Figures S1A–S1C available online). We then examined the interaction of single molecules of Kin5-Dmotor-GFP with surface-immobilized microtubules by using TIRF microscopy (Figure 1B). Previous studies have shown that full-length kinesin-5 has two modes of motion on microtubules, depending on ionic strength of the assay buffer and microtubule binding state [8, 9]. On single microtubules, plus-end-directed, ATP-dependent directional motion predominates in low-ionic-strength buffers, whereas under more physiological conditions ATP-independent one-dimensional diffusive motion predominates. In physiological ionic strength, kinesin-5’s motility switches from diffusive to directional upon crosslinking a pair of microtubules. Based on these results, we mainly used two different buffer conditions throughout this study: one with low salt and one with higher salt that is closer to physiological ionic strength (see Experimental Procedures for details). In the high-salt buffer we did not detect Kin5-Dmotor-GFP on microtubules (data not shown). However, upon reducing the salt concentration, we observed Kin5-Dmotor-GFP binding to bundles of microtubules (Figure 1C). Although multiple binding events could be observed on any one bundle, the association times were very short (<1 s) under our imaging conditions. We could only estimate an upper limit of 1 s for the average association time of this construct with microtubules, which is significantly lower than the previously determined value for full-length Kin5-GFP of tavg = 34 s [8]. Because individual molecules of Kin5-Dmotor-GFP interact briefly with microtubules, we examined whether multiple molecules of this protein could establish more persistent filament interactions (Figure 1E). For this experiment, the glass surface of an assay chamber was coated with Kin5-Dmotor-GFP and blocked to prevent nonspecific attachment of microtubules. When microtubules were added, they bound to the Kin5-Dmotor-GFP on the surface and did not release over a range of salt concentrations. Strikingly, while remaining attached to the surface, each microtubule moved back and forth along its long axis (Movie S1 and Movie S2 available online). We used kymography to track the motion of these filaments (Figures 1F and 1G) and performed mean squared displacement (MSD) analysis of filament positions (Figure 1H). The MSD of the filaments was linearly proportional to the time interval, consistent with one-dimensional diffusion on the surface. Furthermore, the diffusion constant was sensitive to ionic strength, increasing 5-fold from D = 450 6 40 nm2/s in low salt to D = 2600 6 200 nm2/s in high salt. These values are similar in magnitude to those measured for one-dimensional diffusion of microtubules along a surface coated with MCAK, a member of the kinesin-13 family [10, 11]. These data suggest that a microtubule binding site outside the motor domain, most probably in the kinesin-5 tail, allows kinesin-5 to bind microtubules in a way that tolerates slippage along the filament’s axis.

Summary Kinesin-5, a widely conserved motor protein required for assembly of the bipolar mitotic spindle in eukaryotes, forms homotetramers with two pairs of motor domains positioned at opposite ends of a dumbbell-shaped molecule [1–3]. It has long been assumed that this configuration of motor domains is the basis of kinesin-5’s ability to drive relative sliding of microtubules [2, 4, 5]. Recently, it was suggested that in addition to the N-terminal motor domain, kinesin-5 also has a nonmotor microtubule binding site in its C terminus [6]. However, it is not known how the nonmotor domain contributes to motor activity, or how a kinesin-5 tetramer utilizes a combination of four motor and four nonmotor microtubule binding sites for its microtubule organizing functions. Here we show, in single molecule assays, that kinesin-5 homotetramers require the nonmotor C terminus for crosslinking and relative sliding of two microtubules. Remarkably, this domain enhances kinesin-5’s microtubule binding without substantially reducing motor activity. Our results suggest that tetramerization of kinesin-5’s low-processivity motor domains is not sufficient for microtubule sliding because the motor domains alone are unlikely to maintain persistent microtubule crosslinks. Rather, kinesin-5 utilizes nonmotor microtubule binding sites to tune its microtubule attachment dynamics, enabling it to efficiently align and sort microtubules during metaphase spindle assembly and function. Results Kinesin-5 Contains an Additional Nonmotor Microtubule Binding Site Full-length kinesin-5 is a homotetrameric, bipolar protein comprised of three domains (Figure 1A). The motor domain is at the N terminus, followed by the central stalk required for dimerization of motor pairs (coil 1) and tetramerization of kinesin-5 (coils 2–4). At the C terminus, there is an additional 140 amino acid conserved tail domain whose function is unclear. To characterize the function of the nonmotor tail, we generated a fluorescently labeled truncation construct of Xenopus laevis kinesin-5 (named Kin5-Dmotor-GFP) that has the entire conserved kinesin motor domain deleted but includes all of kinesin-5’s nonmotor domains, similar to a Drosophila melanogaster kinesin-5 construct shown previously to crosslink microtubules [6, 7]. GFP was fused to the C terminus, a site where it is unlikely to be functionally

*Correspondence: kapoor@mail.rockefeller.edu

A Nonmotor Microtubule Binding Site in Kinesin-5 Allows One-Dimensional Diffusion of Microtubules (A) Domain structures of kinesin-5 GFP-fusion constructs used in this study. we deleted the entire tail region. a 20-fold excess over concentrations of Kin5-GFP that bundled microtubules (data not shown). Rhodamine-labeled microtubules were immobilized on a glass surface and the interaction of Kin5-Dmotor-GFP was observed by single-molecule TIRF microscopy.6 6 1. Fits represent MSD = 2Dt. bottom. which is atypical for microtubule binding proteins [12]. 60 s. (F and G) Kymographs show the movement of microtubules attached to a glass surface by Kin5-Dmotor-GFP in low-salt buffer (F) and high-salt buffer (G) Scale bars represent 2 mm. To identify the binding site within the tail. (D) Kymograph shows Kin5-Dmotor-GFP interactions with microtubules. which are common in microtubule binding domains. time average over 300 frames). Strong conservation in portions of the C-terminal region was found. In the same buffer conditions. Surprisingly. we utilized an in vitro microtubule sliding assay described before [8] (Figure 2A). single frame. The Kinesin-5 Tail Domain Is Important for Crosslinking Microtubules To examine how a nonmotor microtubule binding site in the C-terminal tail impacts kinesin-5’s ability to crosslink and slide microtubules apart.Kinesin-5 Nonmotor Microtubule Binding Site 155 Figure 1.4 nm/s (n = 20). (E) Schematic of the surface binding assay. This construct (named Kin5-Dtail-GFP). including that of Xenopus laevis. Scale bar represents 4 mm. In high-salt buffer. the tails of many of the kinesin-5 sequences analyzed. we aligned kinesin-5 sequences from 17 species. had a gel filtration elution volume and single-particle fluorescence intensity distribution very similar to those of Kin5-GFP. Further. Movie S3) [8]. error bars = SEM. Because we were unable to identify a specific sequence in the kinesin-5 tail that was likely to be responsible for microtubule binding. See also Figure S1 and Movie S1 and Movie S2. had predicted net-negative charges. To examine relative filament sliding. Kin5-GFP robustly crosslinked microtubules from the solution to surface-immobilized microtubules and drove their relative sliding at 45. (H) MSD calculated from motion of surface-adhered microtubules in high-salt buffer (closed circle) and low-salt buffer (open circle). but no clusters of basic residues. . Rhodamine-labeled. center. (B) Schematic of the single-molecule microtubule interaction assay. and a reaction mix containing motor protein and additional nonbiotinylated microtubules was added. were identified (Figure S2). (C) Images showing Kin5-Dmotor-GFP binding to surface-immobilized microtubules (top. we did not observe microtubule bundling in solution by Kin5-Dtail-GFP at concentrations up to 100 nM. as expected (Figure 2B. microtubule. biotinylated microtubules were immobilized on glass coverslips through biotin-streptavidin linkages. Scale bar represents 4 mm. indicating that this construct is also homotetrameric (Figures S1A–S1C). we designed a kinesin-5 construct in which it was deleted. we did not observe any microtubule crosslinking or sliding by Kin5Dtail-GFP over a wide range of protein concentrations (data not shown). like Kin5-Dmotor-GFP.

Significantly. reflecting tetrameric motors.Current Biology Vol 21 No 2 156 was not tagged with GFP to generate microtubule sliding pairs. and Kin5-Dtail-GFP (green diamonds). In low-salt buffer. At low ionic strength.05 mM. high concentrations of Kin5Dtail-GFP were capable of microtubule bundling in solution (data not shown). we used full-length kinesin-5 that . and accumulated in bright aggregates at the microtubule tips (Figure 2D. but the average sliding velocity was 21. Scale bar represents 3 mm. The Kinesin-5 C-Terminal Tail Domain Is Required for Relative Microtubule Sliding at Physiological Ionic Strength (A) Schematic of the microtubule-sliding assay.4 and 2. which drives relative sliding. consistent with the motion being driven by kinesin-5 molecules walking on one microtubule. limited Kin5-Dtail-GFP binding was observed along the microtubules (Figure 2C. Together. The intensities of moving fluorescent spots. Dashed red lines in center kymographs indicate position of moving microtubule. This suggests that the aggregates at the microtubule tips may be responsible for the infrequent sliding events observed. Error bars indicate SEM. In contrast. Rhodamine-labeled microtubules from solution were crosslinked to surface-immobilized microtubules by kinesin-5. Figures 3C and 3H). In an elevated ionic strength buffer (PEM70 + 40 mM KCl). and (D) Kin5-Dtail-GFP (3 nM) in low-salt buffer. moved toward the plus ends.0 3 103 nm2/s. consistent with previously published results [8]. and Kin5-DmotorGFP by using cosedimentation assays (Figure 2E). bound with more than 10-fold weaker affinity (Kd = 2. respectively). (E) Microtubule binding curves for Kin5-GFP (red circles). and MSD analysis was not feasible.7 nm/s. rather than on both simultaneously. Kin5-GFP made long associations with microtubules (average association time = 29 s.19 6 0.9 6 1.7-fold greater than that of Figure 2. Microtubule affinity was determined by cosedimentation of the kinesin-5 constructs with microtubules in low-salt buffer with 2 mM ADP. green arrows in kymograph). Rare crosslinking and sliding events were observed (Figure 2D. green). with a Kd of 0. and Movie S5.5 mM. and binding constants were determined by fitting to a hyperbola. The Kinesin-5 Tail Domain Promotes Microtubule Association and Processivity We next examined the effect of the C-terminal tail deletion on the motility of individual Kin5-Dtail-GFP molecules on single microtubules. The movement of Kin5-Dtail-GFP molecules along surface immobilized microtubules was recorded in various salt conditions and compared to that of Kin5-GFP (Figures 3B–3G). Microtubule sliding in the presence of (B) Kin5-GFP (1 nM) in high-salt buffer. under the same conditions. Kin5-Dtail-GFP and Kin5-Dmotor-GFP. Full-length kinesin-5 bound tightest. Average microtubule association times were determined by fitting histograms of individual run durations to single exponential functions (Figures 3K–3M). We compared the microtubule binding affinities of full-length Kin5-GFP. Movie S5). and the motility was mostly diffusive (v = 0.3 nm/s (n = 11). We next examined the effect of ionic strength on Kin5Dtail-GFP binding of microtubule pairs. both constructs moved processively and directionally along the microtubule (Figures 3B and 3E). Figure 3K). Kin5-Dtail-GFP clearly decorated immobilized microtubules. but can crosslink preorganized antiparallel filaments. Figures 3L and 3M). Deletion of the C-terminal tail in Kin5-Dtail-GFP could cause microtubule crosslinking to fail because of reduced microtubule affinity. Kin5-Dtail-GFP. See also Figure S2 and Movie S3.1 6 0. each of the truncation constructs. This indicates that both the motor and tail domains of kinesin-5 contribute to the overall binding of full-length kinesin-5 to microtubules. To examine this. though the interactions of Kin5-Dtail-GFP with microtubules were generally shorter than those of full-length kinesin-5 (average association times = 29 s for Kin5-GFP and 11 s for Kin5-Dtail-GFP. Kin5-Dtail-GFP was unable to make sustained interactions with microtubules (Figure 3F). (C) unlabeled kinesin-5 (2 nM) and Kin5-Dtail-GFP (3 nM) in high-salt buffer.4 6 0. (B–D) Kymographs show the motion of microtubules (left. Additionally. For this we used a single-molecule TIRF microscopy-based motility assay (Figure 3A). It is possible that Kin5-Dtail-GFP cannot align microtubules on its own. The fraction bound was calculated from the relative fraction remaining in the pellet after cosedimentation. Green arrows indicate Kin5-Dtail-GFP accumulated at microtubule tips. Kin5-Dtail-GFP (blue squares). D = 4. were fit by 2D Gaussians and single-particle tracking was used to determine trajectories of the individual motors for mean squared displacement analysis (Figures 3H–3J). Movie S4. these results suggest that the C-terminal tail domain is required for kinesin-5’s ability to crosslink microtubules and drive their relative sliding. red) and Kin5-GFP constructs (center. Movie S4). Even under these conditions with relatively high concentrations of Kin5-Dtail-GFP. MSD analysis showed that the average velocity of Kin5-Dtail-GFP was about 1.

4 6 0. (K–M) Histograms of association times for Kin5-GFP in PEM70+40 mM KCl (K) and for Kin5-GFP (L) and Kin5-DtailGFP (M) in PEM70. Solid lines are fits to MSD = v2t2 + 2Dt + offset. Average association times.7 6 0. Kin5-GFP (14. (B–G) Kymographs showing the motion of (B–D) Kin5-GFP and (E–G) Kin5-Dtail-GFP along microtubules in PEM70 (2 mM ATP) and increasing concentrations of KCl. We compared the motility of Kin5-Dtail-GFP with those of Kin5-GFP and Kin5-dimer-GFP.5 nm/s for Kin5-GFP (n = 114 microtubules from 4 independent experiments). determined from fits to single exponentials (solid curves). Kin5-GFP’s velocity. such that the velocities of the two tetrameric kinesin-5 constructs were nearly identical. In low-salt buffer. to v = 35.1 and 8. with v = 29. 1500 pM in PEM70+80 mM KCl (G).7 nm/s for Kin5dimer-GFP (n = 85 microtubules from 3 independent experiments). Taken together.3 3 103 nm2/s and D = 1. consistent with a reduced microtubule binding affinity under these conditions. respectively. however. In low salt the . Figures 3J and 3I). respectively) suggesting that diffusive motion can be a property of the kinesin-5 motor and that the presence of the tail can suppress diffusion in low-salt conditions.1 6 0. microtubule binding and gliding was observed for all three kinesin-5 constructs (Figures 4B–4D). Kin5-GFP.2 6 0. are indicated. 120 pM in PEM70+40 mM KCl (C). [Kin5-DtailGFP]: 150 pM in PEM70 (E). Because motorless kinesin-5 showed saltdependent interactions with microtubules (Figures 1F–1H). Kin5-Dtail-GFP also exhibited a slightly higher diffusion constant than Kin5-GFP (D = 2. Scale bars represent 3 mm. Singlemolecule TIRF microscopy was used to observe the motion of individual kinesin-5 molecules moving along surfaceimmobilized microtubules. had about 1. 600 pM in PEM70+40 mM KCl (F).8 6 0. error bars indicate SEM. increased by more than 2-fold.4 nm/s for Kin5-DtailGFP (n = 84 microtubules from 3 independent experiments) and v = 30.Kinesin-5 Nonmotor Microtubule Binding Site 157 Figure 3. The average gliding velocity of the two truncation constructs were similar. In high-salt buffer. and the movement of microtubules along the surface was recorded by time-lapse fluorescence microscopy. consistent with the results of the single-molecule assay. indicating that the tail domain can modestly attenuate kinesin-5 motility. Motor protein concentration had to be increased to obtain sufficient numbers of binding events as ionic strength increased. we employed a microtubule surface gliding assay to examine Kin5-Dtail-GFP’s motor activity (Figure 4A). microtubules rapidly dissociated from the Kin5-dimer-GFP-coated surface at all protein concentrations tested. the average gliding velocity of Kin5-Dtail-GFP increased slightly. (H–J) Mean squared displacement (MSD) analysis of motility on single microtubules for Kin5-GFP in PEM70+40 mM KCl (H) and for Kin5-GFP (I) and Kin5-Dtail-GFP (J) in PEM70. The Kinesin-5 Tail Domain Does Not Interfere with Motility To further probe the effect of the C-terminal domain on kinesin-5 motility. and 240 pM in PEM70+80 mM KCl (D). In the high-salt buffer. Glass coverslips were coated with different kinesin-5 constructs.4 3 103 nm2/s. to v = 34. however. these results indicate the kinesin-5 nonmotor tail domain enhances microtubule association but has a modest. salt-sensitive effect on motility.5 nm/s. [Kin5-GFP]: 60 pM in PEM70 (B).4 nm/s (n = 84 microtubules from 3 independent experiments). The Kinesin-5 Tail Domain Increases Processivity and Association Time with Microtubules (A) Schematic of the single-molecule motility assay. a dimeric kinesin-5 truncation construct described earlier [8].7-fold lower average velocity (Figure 4E). we next examined whether the tail deletion affects the salt sensitivity of kinesin-5 motility. with v = 16.6 nm/s (n = 136 microtubules from 4 independent experiments).

the motor and nonmotor domains mediate an interaction with the filament in which it samples multiple orientations. A requirement for nonmotor microtubule binding sites was largely unexpected for kinesin-5. and Kin5-dimer-GFP (D) in low-salt buffer (left) and high-salt buffer (right). the tail domain increases kinesin-5’s microtubule association without substantially resisting motility. When a kinesin-5 molecule in solution collides with a microtubule.Current Biology Vol 21 No 2 158 Figure 4. align. and that kinesin-5 uses these motor and nonmotor domains to tune its microtubule interactions so individual kinesin-5 molecules can efficiently crosslink. but this impedance disappears as ionic strength is increased. The inset highlights how the nonmotor tail domains maintain association with the filament and prevent full dissociation of kinesin-5 when the low-processivity motor domains unbind. (ii) The tail domain mediates the initial interaction. X-rhodamine-labeled microtubules bind to kinesin-5 constructs coating a glass surface. it is difficult to rule out the alternative mechanism in which the tail domain may interact with the motor domain to allosterically regulate its microtubule affinity and motor activity. Although we favor a simple model wherein the primary function of the tail is to mediate direct interactions with microtubules. allowing it to dwell on the filament and explore its length via 1-D diffusion. (E) Average velocities for each construct in two buffer conditions. The presence of the C-terminal tail domain is also required to crosslink and slide two microtubules apart. (F) Model for the contribution of the microtubule binding site in the C-terminal tail domain to kinesin-5 motility. Discussion Our findings suggest that tetramerization of the kinesin-5 motor domains alone is not sufficient for microtubule crosslinking. (i) A kinesin-5 molecule in solution prior to encountering a microtubule. The Kinesin-5 C-Terminal Tail Domain Reduces Motility in Low Ionic Strength (A) Schematic of the microtubule surface motility assay. Instead. our data indicate that kinesin-5 requires a combination of eight microtubule binding sites (four nonmotor and two pairs of low-processivity motor domains). Interestingly. Based on our findings we propose a model for how kinesin-5 homotetramers can walk on each filament it crosslinks (Figure 4F). presence of the tail in Kin5-GFP moderately reduces motor activity. This prolonged interaction with one filament would increase the likelihood of interacting with a second. Kin5-Dtail-GFP (C). (iii) Relative filament sliding driven by kinesin-5’s two pairs of motor domains. in part because it appeared that two pairs of motor microtubule binding sites would be sufficient for microtubule crosslinking. allowing 1-D diffusion. and slide apart microtubule pairs. The microtubule velocity in each kymograph was determined from the slope of the red dashed line. The population average for Kin5-dimer-GFP in high salt was not determined because the microtubules rapidly detached from the surface. and increases the probability of crosslinking a second microtubule. Once a second microtubule makes . Error bars indicate SEM. (B–D) Examples of kymographs showing the movement of microtubules driven by Kin5-GFP (B).

Velocity and diffusion constant were determined from MSD analysis and fitting as in [8]. suggesting that monastrol stabilizes kinesin-5 in a low-friction microtubule-attached state [9. and an additional nonmotor microtubule binding site that passively binds a second filament and carries it as cargo.8 with KOH]). The low processivity of the motor domains. would allow kinesin-5 to efficiently slide microtubules apart without having to form large multiprotein ensembles. Moving fluorescence spots were detected as in [27] and their trajectories were recovered by single-particle tracking [28] along microtubules imaged in advance. and the motor pairs at each end of the dumbbell can move toward the microtubule plus ends. many kinesin-5 inhibitors that have recently entered clinical trials as cancer therapeutic agents inhibit kinesin-5 by stabilizing the ADP-bound state of the motor domain. probably cannot maintain microtubule crosslinks when the motors dissociate. A large fraction of the kinesin-5 in this region is positionally stable despite microtubule flux toward the poles. whether the two microtubules are oriented parallel or antiparallel. Error bars in figures indicate SEM. Motility Assays Photobleaching and motility assays were performed on an inverted microscope (Axiovert. where antiparallel microtubule overlap is most prevalent. but in the opposite direction. 24]. Kinesin-5 localizes to the center of the spindle. kinesin-5 has been considered as a potential drug target. MSD Analysis Motility data were analyzed with customized software written in MATLAB (Mathworks). where microtubule minus ends are clustered and many filaments are parallel [15]. monastrol treatment of cells blocks spindle pole separation but does not eliminate localization of kinesin-5 to the monopolar spindles [25]. Because of its essential role in cell division. disruption of this interaction may represent a good strategy for designing highly selective anticancer drugs. it has been shown that a 120-fold stoichiometric excess of kinesin-14 is needed to resist the oppositely directed microtubule-sliding activity of kinesin-5 [20]. Carl Zeiss) as described previously [8. The low-salt buffer was PEM70 (70 mM Pipes. Individual directional runs are likely to be short because kinesin-5 motors have low processivity and tend to dissociate from microtubules after walking w67 nm [13. each of which has a single pair of motor domains that walks on one filament. Therefore. which involve motor and nonmotor domains. such as kinesin-1 and the minus-end-directed motor kinesin-14 [16–18]. with some modifications.Kinesin-5 Nonmotor Microtubule Binding Site 159 contact at the opposite end of the kinesin-5 tetramer. whose motors are highly processive. faster transport mechanisms driven by dynein. The interaction between the kinesin-5 nonmotor domain and microtubules is essential for microtubule-sliding function and is a feature unique to kinesin-5. According to our model. when a kinesin-5 molecule uses its two pairs of motor domains to walk along two microtubules. whose motor domains are nonprocessive [19]. 1 mM MgCl2 [pH 6. such as chromokinesins. the nonmotor domains would allow kinesin-5 to maintain the crosslink without encumbering continued sliding driven by other molecules of kinesin-5. and the formation of specialized microtubule bundles. kinesin-14 probably requires a large number of molecules to sustain microtubule crosslinking. Nonlinear least square fitting was applied on original MSD data with the relation MSD = v2t2+2Dt+offset. Our model can help explain the two main features of kinesin5’s distribution in the metaphase spindle. Experimental Procedures Protein Constructs See Supplemental Information for details. combined with low-friction microtubule interactions via the nonmotor domains. then kinesin-5 would be pulled along by the motion of the microtubules. Kinesin-5’s walking speed along microtubules is comparable to the rate of microtubule flux. could enable kinesin-5 to act as a slipclutch. Kinesin-14. A single molecule of kinesin-1. when this occurs. thus locking the motor in a conformation that has low affinity for microtubules [21–23]. but not ATP-independent 1-D diffusion on microtubules. However. microtubule association can be maintained by the binding sites in the C terminus of the full-length homotetramer. . the interactions with microtubules. X-rhodamine-labeled. Microtubule Sliding For the microtubule relative sliding experiments. See Supplemental Information for details. 14]. See Supplemental Information for details. the motor protein’s fixed position would still be maintained by the motility of the second pair along the other filament. 1 mM EGTA. biotinylated microtubules were immobilized on a glass surface with a ‘‘biotin sandwich. the nonmotor domains would keep kinesin-5 associated with both filaments it crosslinks despite frequent stochastic dissociation of the motor domains. However. explaining the observed relative enrichment of kinesin-5 on microtubules near the spindle poles. Significantly. The eight binding site mechanism also has implications for how kinesin-5 interacts with other mechanical components of the spindle.’’ See Supplemental Information for details. When both pairs of motor domains disengage simultaneously and the crosslink is maintained solely by the nonmotor domains. Between parallel microtubules the motor protein would ride polewards on both. directional motility is triggered. Therefore. Kinesin-5 also accumulates at the spindle poles. the low-processivity motor domains tend to dissociate frequently from microtubules. Like monastrol. The proposed mechanism would be distinct from those of other motor proteins that can drive relative microtubule sliding. but probably would not be compliant with the functions of other cytoskeletal components as is predicted for kinesin-5. In fact. Our findings explain these observations because although monastrol reduces the microtubule affinity of the motor domains. 26]. such as the kinetochore fibers that link chromosomes to the spindle. Thus kinesin-5 could accommodate filament motion directed by other motor enzymes. 9. When one pair of a kinesin-5 tetramer’s motor domains disengages from one microtubule. such that some fraction of the molecules present between two filaments have their motor domains bound. Between antiparallel microtubules the motion would be balanced in both directions and the kinesin-5 molecule would have little net movement. Therefore. Cosedimentation Microtubule binding assays were performed essentially described [26] in PEM70 in the presence of 2 mM MgADP. The high-salt buffer was PEM70 with 80 mM KCl added. it can maintain an approximately fixed position relative to the spindle. Two different buffers were used for most motility experiments. Together. is likely to make sustained microtubule crosslinks. Inhibition of full-length kinesin-5 by monastrol abolishes directional motility. with the ability to disengage its motors but maintain microtubule crosslinks in situations where its motor activity is overwhelmed by the function of other spindle components.

Coy. Cross. Biochemistry 39.. IEEE Comp. Zhao. 114–118.M. G.D. Nature 441. Civelekoglu-Scholey.. (2009).S. W. (2009). 12151–12156. Luo. Lakonishok. The mitotic kinesin-14 Ncd drives directional microtubule-microtubule sliding. (2003). and Diez.M. D. DeBonis.. Hajdo.M.. Cell Biol. 1784–1791... 338–349. and Cross. Kapitein. Schmidt. Proc.. C. P. Hunter. Interaction of the mitotic inhibitor monastrol with human kinesin Eg5. Waterman-Storer. L. (2010). Nat. Chem. 24.J. Jolly..K. J. 21. Kapoor. Soci. 2011 References 1.L. Curr.H.M.. H.M.. Krzysiak. Eg5 is static in bipolar spindles relative to tubulin: evidence for a static spindle matrix.. 2.. Hancock. 115–119. 3336–3352. T.A.D. 16. A. Computational analysis of F-actin turnover in cortical actin meshworks using fluorescent speckle microscopy. Kapitein. Gans.Current Biology Vol 21 No 2 160 Supplemental Information Supplemental Information includes Supplemental Experimental Procedures. (2000). A.J. and Gilbert. R. R. KLP61F.. A. Cell Biol. Shaw.. S. van den Wildenberg. K. (2003). and Peterman. 23. S. Crevel. Nature 379. Biol. 182.. Acad. Microtubule organization by the antagonistic mitotic motors kinesin-5 and kinesin-14. L. J..S. Mol.H.038. (2005). Cell Biol. T. Skowronek.. 22913–22916. J. L. Kim. A. T. C. S. A homotetrameric kinesin-5.G. Probing spindle assembly mechanisms with monastrol..M. cross-links microtubules within interpolar microtubule bundles of Drosophila embryonic mitotic spindles.H. V. Evidence that monastrol is an allosteric inhibitor of the mitotic kinesin Eg5.. 19. 11.. Crevel. (2010).. Lockhart. and Howard. J. and National Science Foundation (NSF) grant MCB-1052660 to G..L. 3.M.M. L. Kim. J. L. Lehr... M. and Schmidt... E.... 465–480.. G.G. G. C. Scholey. J. 22. W. C. Allosteric inhibition of kinesin-5 modulates its processive directional motility.. Curr. K.C. E. Received: November 16. Reliable tracking of large scale dense antiparallel particle motion for fluorescence live cell imaging..N. D. K. McDonald.K. C. Tao. was supported by National Research Service Award postdoctoral fellowship (F32 GM083566) and an Institutional Research Training Grant (T32 CA09673) from the NIH. K. S. Kwok. Carson. and Scholey. G.. Chem.C. (2002). A bipolar kinesin. Evans.M. V.J. T.M. J. Biol.. G. J. Biol. A.M. A. M. L. 7..... Weinger.D. Kwok. Caplow. 445–457. Valentine. Tao. Peterman. Kapitein.M. 25. 189. J.. T. C.. B..E. Sheehan. 154. two figures. Cell Biol.. B.. Foster.A. 144... Yang.L. L.. Curr. Natl. Schmidt.J.. 18. Saxton. Skoufias.H.M. (1994).. J.P... Biol.. Kapoor. Kim. T. and Mitchison.M. Brown. (2006). D. Nature 359.. Cell Biol. J.. J.A.W. R. Ponti. Drummond. Cooper.M. 2. G. bundles microtubules and antagonizes Ncd in motility assays. (2004).. C.S. M. A.G.. 5. Sci. Kapoor. LeGuellec.J. Maliga.J. USA 107. and Kapoor. Biol. Kalaidzidis. 15.J.. Mitotic spindle organization by a plus-end-directed microtubule motor. L.. P.... (1992)... Diez. Wollman. and Peterman. T. and Howard. and Mitchison. 1783–1788. Curr. and Danuser. and Block.F.J.I. C. 68–73.W. 722–726. and Danuser.. (1996).U. Chem. ATP-competitive inhibitors of the mitotic kinesin KSP that function via an allosteric mechanism.. R. Helenius. et al. Blangy. Coughlin. Z. a small molecule inhibitor of the mitotic kinesin. 270–272. 18. 1860–1864. Cell Biol.. T.M. 20. 421–428. C..M.. Mol. E. 724–730. 717–723. A. (2007). T. Conf. (2006).. KLP61F. 150.R. The depolymerizing kinesin MCAK uses lattice diffusion to rapidly target microtubule ends. and Wordeman. Matthies. Cell 11.C.. Stahlberg.M. and five movies and can be found with this article online at doi:10. J.M. 28. L... H. Sawin. Yang. 12. Sharp. and Gelfand. and McAinsh.. The kinesin-related protein MCAK is a microtubule depolymerase that forms an ATP-hydrolyzing complex at microtubule ends. Fordyce.. 269. 21. 3. Biol.. Biol. R. Kapoor. J.. Kwok. 540–543.P. and Mitchison. Cross. 26. 8.. A. Alonso. E. 138.. D. J. Nevins. (2008). (2005).. 17.. Braun. 160–170. Acknowledgments We thank members of the T. D. K.. Nat. (2008)..L. Sudakin.T. P. D. H. Hentrich. J. 84. A ‘‘slow’’ homotetrameric kinesin-related motor protein purified from Drosophila embryos.. Parrish. et al. lab for helpful discussions. Kinesin-1 heavy chain mediates microtubule sliding to drive changes in cell shape. 13. R411–R412. . Fink. Walczak.P.M.A. A. Chem. The bipolar mitotic kinesin Eg5 moves on both microtubules that it crosslinks. T.A. B. Hackney..F. Biophys. Mitchison.. 11. Biol. The diffusive interaction of microtubule binding proteins. was supported by an ICES-DOWD Fellowship. Saxton. 8. Reuther.Y. CVPR 3.. Nat. A. A. Gilbert.M. 125–138.C. Mayer. Vale. Biochemistry 42. Proc. 470–476. and Scholey. Srinivasan.M. K. 989–996. 10. (2000). J. and Scholey.. J. D. H. Simorre. T. T. This research was funded by National Institutes of Health (NIH) grant GM065933 to T. I. T. M. 16. Baskin.A... The kinesin-14 Klp2 organizes microtubules into parallel bundles by an ATP-dependent sorting mechanism. 2010 Published online: January 13.1016/j. I... Ebel. 273. Kashina. Biol. (1997).. 2010 Revised: December 14. S.C. R. Individual dimers of the mitotic kinesin motor Eg5 step processively and support substantial loads in vitro. Schmidt.. Opin.. J.G. Microtubule cross-linking triggers the directional motility of kinesin-5. 2010 Accepted: December 16. Nature 435. D.F. S.M. T. (2001).C.Q. W. Curr. J.. Cell Biol.C. Diez. Kapitein. 1125–1133.. 14. 14. C.P. Chaudhari. Wedaman. L.cub. M. M. Kasprzak. J. M. R. K. N. and Cross. H. Cole.J.F. W. I.... Philippe.B. J.H.H. Nat.. 480–485. 975–988.. Y... McNulty. Larson. S. and Mitchison. The bipolar kinesin. and Surrey. (2006). The rate of bipolar spindle assembly depends on the microtubule-gliding velocity of the mitotic kinesin Eg5.. Nat.O. Lebeau. Cole.12. 9.J.J.A. 9.. 2293–2302. Kapoor. The homotetrameric kinesin-5 KLP61F preferentially crosslinks microtubules into antiparallel orientations. T. S. L. J... 27. Mogilner. J. J. Wordeman.M. Peterman. Crevel.E. (2004). (2006). D.. Cell Biol.D.. Cell Biol.. Kinetic evidence for low chemical processivity in ncd and Eg5. Monastrol stabilises an attached low-friction mode of Eg5.R.J. Kwok.. A. 14. Vallotton. B.2010. 6.J. 11. R. Eg5. M. (2009). 4. and Kapoor. (1999)... (2003). and Scholey. Brouhard. Salmon. Kinetic studies of dimeric Ncd: Evidence that Ncd is not processive. Matov.